Abstract
In semi-arid tropical areas, sorghum is sown at very low planting densities. Hence, increasing plant density represents an opportunity to improve productivity. However, assessing the expected increase in water needs is critical prior to testing higher densities under rainfed conditions. This was tested with a panel of elite cultivars in field and lysimiter experiments, and testing the effects of two density treatment, high (HD, 22 plants.m -2 ) and low (LD, 11 plants.m -2 ), on grain and biomass yield and on water use and water use efficiency (WUE). Doubling the conventional sowing density significantly increased biomass and grain yield, with a genotypic variability in the biomass response. No link was found between the response to density and the maintenance of the tillering capacity, whereas the response to density was somewhat explained by a differential increase in the leaf area index under high density (r=0.43 P<0.05). Lysimeter experiments showed that, compared with the conventional density, the high-density treatment had 62% increase in biomass vs a 38% increase in water use, resulting in a 17% higher WUE on average of the genotypes tested. There was an appreciable genotypic variability in this degree of WUE increase under high density. The most striking result was the very tight positive link between the biomass response to density and the differential increase in WUE in the dry season (r=0.91 P<0.0001), whereas in the wet season this link was negative (r=-0.48 P<0.02). This work shows that intensifying sorghum production by increasing sowing density is possible, in the short term using cultivars that show the largest WUE increase under high density, in the longer term by breeding high-density adapted cultivars, targeting plant traits that explain the tight link between higher WUE and higher yield under high density